 |
 |
 |
 |
At 1024 channel resolution, the resulting files are more than twice
as big as for 8-bit data. Not only is there an extra byte per param-
eter but, with 4 times as many channels, 4 times as many events need
to be recorded to give decent histograms with good statistics (de-
pending on the dispersion of the data set). Unless, of course, you
rebin the data down, but then why not just collect 8-bit data in the
first place? Disk space is cheap these days (less than $1/mB for
many good disks) but who wants to wait while your computer grinds
through a 1024 X 1024 channel contour plot?? Log amp transfer
functions aside, momentarily, most of the software I've seen usually
rebins to 64 X 64 channel bivariates or maybe 128 X 128 at most!
On the point of the transfer function - since Parks, et al, used
beads in a flow cytometer, as opposed to pulses from a pulse
generator, isn't the majority of the fluctuation readily attributable
to the cv of the measurement of the fluorescence intensity?
Robb Habbersett
|
|
|
|